Prediction of Computer Type Using Benchmark Scores of Hardware Units
Yavuz Selim Taşpınar, İlkay Çınar, Murat Köklü
Abstract
Users need an expert opinion to learn about their current computer or purchasing. In addition to these, computer and computer component manufacturers have to carry out innovation studies such as improving the products they produce by receiving feedback about the products they produce, and changing the marketing strategy. There is various benchmark software to meet all these needs. This benchmark software measures the software and hardware performance of the computers and enable users to gain information about the performance of their computers and components. The category of computers can also be determined as a result of the performance evaluation obtained. Various statistical and machine learning methods are used to determine these categories. In this study, it is tried to predict which category the computers fall into by using the computer features in a dataset obtained from the internet by the web scraping method by random forest and logistic regression method. The effect of computer features in the dataset on classification has been analyzed. Classification success was 89.4% with the random forest method and 84.3% with the logistic regression method.